Moneycontrol PRO

HomeNewsTechnology startupAI edge newsletter

Stay ahead with the AI Edge

Get the latest AI breakthroughs, industry insights, and technology trends delivered directly to your inbox every Saturday.

Latest Newsletters

AI in India’s fintech: Hype vs reality
Let's be real...everyone wants to put "AI-powered" on their pitch deck right now. But here's the thing, India's fintech players are being cautious about it. And that might be the smartest move they're making. Because when you're dealing with lakhs of crores in transactions and regulators who actually read the fine print, you can't just YOLO your way into artificial intelligence. Gold rush (but on training wheels): At the recently concluded Global Fintech Fest 2025 in Mumbai, AI was everywhere. NVIDIA and the National Payments Corporation of India (NPCI) built an entire "Bharat AI Experience Zone." DCB Bank showed off multilingual AudioBots. Gnani.ai demoed AI-generated digital humans. But most of it? Still in the lab. "Everyone's talking about AI in banking, but very few are actually using it meaningfully," says Sabyasachi Goswami, CEO of Perfios. For newer fintechs, AI is a clean slate, fresh data, small teams, and quick wins. But for legacy players? Even tiny changes are painful, and ROI is a question mark. “There’s a lot of FOMO,” confessed one founder. “Last year, the RBI wasn’t excited about AI. This year, everyone wants to show they’re doing something. But it’s all very experimental.” Ganesh Gopalan of Gnani.ai hit the nail on the head. "You can wait five seconds for ChatGPT. But if your bank app takes that long to load your balance, you'll slam the phone," Gopalan said. Innovation meets regulation: The biggest blocker isn't tech...it is trust. And regulation. "AI adoption in financial services isn't about letting a thousand flowers bloom...You need a clear top-down blueprint. Random experimentation doesn't scale," explains Neetu Chitkara from BCG. Sahil Kini, CEO of RBI Innovation Hub (RBIH), agrees. "Anyone building AI in fintech without reading the RBI's AI report is doing themselves a massive disservice," Kini said. The report lays out seven key principles, including trust, people-first design, and explainability, which together form the textbook for responsible AI. Which basically means: innovate fast, but crash the regulator's party and you're done. Cost-cutter, yes, but not a cash cow (yet): Right now, most fintechs are using AI as a back-office upgrade. Automating calls. Writing memos. Summarising data. Useful? Sure. Revolutionary? Not so much. A few companies are using AI for smarter recommendations or voice assistants to improve customer engagement. An even smaller number are creating completely new products powered by AI. “The next frontier,” says Chitkara of BCG, “is when AI helps firms cross-sell smarter or build entirely new digital offerings.” Big players are making moves: While most test the waters, some are diving in headfirst. Paytm recently unveiled an AI-powered POS that talks to merchants in local languages. "Every founder needs one teammate; this AI will be that,” Vijay Shekhar Sharma said. Razorpay, NPCI, and OpenAI are taking it to the next level. Agentic Payments inside ChatGPT. Imagine searching for products, comparing prices, and paying via UPI, all without leaving the chat. Investors are paying attention: Z47's Vikram Vaidyanathan says a third of the firm's new investments are in AI-led fintechs. "Earlier, trust came from a bank branch. Today, phenomenal digital experiences build trust." TVS Capital’s Gopal Srinivasan sees AI as “a democratising force,” with value in the foundational tools, the “picks and shovels,” that power financial innovation. The long game: AI won't flip India's fintech world overnight. It'll creep in through credit models, fraud detection, voice bots, smarter risk scoring, quietly, deliberately, and always with one eye on compliance. While global tech sprints toward AGI and sentient chatbots, India's fintech ecosystem is running a marathon. Because in finance, slow and sure doesn't just win the race, it keeps you in business. Read the full story Was this newsletter forwarded to you? You can sign up for the AI Edge here
November 1, 2025
Why Quantum and AI together are a cybersecurity nightmare
Remember when we thought AI-powered cyber threat detection was the endgame? Turns out, we were bringing a neural network to a quantum fight. While enterprises are busy deploying AI to catch hackers faster, a much scarier arms race is brewing in the background, one where quantum computing could crack every encryption standard we rely on and AI might be the accelerant that makes it all happen sooner than we think. The exponential collision no one is ready for: “Think of it as two exponential technologies colliding,” Saugat Sindhu, Global Head of Advisory Services in Cybersecurity & Risk Services at Wipro, told us. “The intersection of Generative AI (Gen AI) and quantum computing will be tenfold more dangerous because of the processing power quantum brings.” Translation: AI makes attacks smarter. Quantum makes them unstoppable. Together? They're rewriting the entire playbook The first thing to break? Encryption, the mathematical lock protecting your bank account, your government's secrets, and basically everything that happens online. The wake-up call: Wipro's latest cybersecurity report doesn't mince words: 75% of enterprises lack the expertise to defend against AI-led attacks, while 86% of recent breaches involved nation-state actors. Sindhu says the fix requires a two-front war. "It's a two-pronged model involving infrastructure hardening around post-quantum encryption and AI model hardening against adversarial attacks like evasion and data poisoning," he explained. In other words, you need to quantum-proof your systems and teach your AI not to get fooled. VCs smell opportunity: Where there's existential risk, there's venture capital. Accel's Prayank Swaroop sees this as a greenfield moment for founders. "If quantum becomes real, existing cybersecurity solutions will fail…That again gives a large surface area for Indian founders to go after. Some founders should start investing time and energy toward it," Swaroop told us. The message is clear: if you're building in security, now's the time to get quantum-serious. Enter Agentic AI: Coforge's Chief Technology Officer Vic Gupta is betting on "opinionated intelligence" as the next line of defence. "Agentic AI will be central to cyber defence in the coming years," he said. "Our goal is to ensure AI can autonomously detect, respond, and remediate without losing control of its boundaries." But here's the catch: autonomy cuts both ways. Global tech research firm Forrester predicts at least one major public breach caused by Agentic AI within two years, as autonomous systems begin making real-time decisions in production. Security teams, it warns, must secure AI intent and track data origin to avoid cascading failures. Quantum rewrites the rules: Forrester expects quantum-safe spending to exceed 5% of global IT security budgets by 2026. Governments are moving fast, the US and EU are funding quantum migration programmes, while India is studying its impact on data sovereignty and defence systems The clock is ticking. Even if large-scale quantum computers are years away, the data being stolen today could be decrypted tomorrow once quantum arrives. It's called "harvest now, decrypt later," and it's already happening. The automation paradox: Gartner's Cybersecurity and Risk Outlook 2026 highlights the irony: AI will make systems smarter and faster, but overreliance on automation without human oversight could create new systemic risks. As Sindhu put it: “Threat modelling must now extend beyond networks and data to AI systems themselves.” The bottom line: The future of cybersecurity will not be code versus code. It will be compute versus compute. Quantum may break encryption, but AI will decide who holds the master key. Dig deeper Was this newsletter forwarded to you? You can sign up for the AI Edge here
October 25, 2025
Powering India’s AI Mission: The energy question
Here's the thing about training world-class AI models: they're incredibly thirsty. Not for data, well, yes for data, but for electricity. India just got a front-row seat to this reality when Google dropped $15 billion on an AI hub in Visakhapatnam. It's Google's biggest investment outside the US, and it's about to stress-test everything we thought we knew about India's energy infrastructure. Clusters, GPUs, and training runs: India's IndiaAI Mission isn't playing small. The initiative has already deployed over 38,000 GPUs across national compute clusters, exceeding its original 10,000-GPU target. For context, training something like GPT-4 ate up around 50 gigawatt-hours in a single run India's taking a different approach, distributing compute across multiple clusters. “If GPT-4 training consumed ~50 GWh in one continuous run, then each India AI GPU cluster training a similar model may consume 30–70 GWh (including overheads). Ten such clusters would amount to ~300–700 GWh annually, but since they run episodically (weeks or months), the average load shall be much lower, and with intelligent scheduling tied to renewables, the energy burden shall be manageable,” Sunil Gupta, co-founder and CEO of Yotta Data Services told us. AI workloads currently represent less than 10% of India's total data centre power use. But they're doubling every 12–18 months, Gupta added. How hungry is AI, really?: By 2030, India's data centre load is expected to jump from 1.2 GW today to 4.5 GW. AI-driven facilities alone could add another 40–50 terawatt-hours annually. To put that in perspective: a typical AI data centre uses as much electricity as 100,000 households. The biggest ones under construction? Try 20 times that. As of July 2025, India's total installed capacity stood at 490 GW. Fossil fuels account for 49.7%, non-fossil sources 50.3%. The country hit its 50% renewable target five years ahead of schedule, but the AI boom is about to test that achievement hard. Electricity is only half the battle. AI workloads generate 70–150 kilowatts of heat per rack—far beyond what traditional data centre cooling can handle. And in a water-stressed country like India, the conventional solution (more water cooling) isn't really an option. India’s data centre landscape: India now has 268 operational data centres, ranking eighth globally. But the real action is in the pipeline: Reliance Industries: Building a 3 GW AI facility in Jamnagar, Gujarat, among the world's largest Tata Consultancy Services: Planning 1 GW of AI-grade capacity over 5-7 years OpenAI: Reportedly exploring partnerships with Indian data centre firms to localise its $500 billion Stargate project Andhra Pradesh's pitch: No state is betting harder on AI infrastructure than Andhra Pradesh. State IT Minister Nara Lokesh sees cheap renewable energy and surplus Godavari water as strategic advantages. “We've promoted renewable energies at a big scale. At a gigawatt scale, we've signed with Tatas, we've signed with Renew. GreenCo is a homegrown company and it's not just about solar and wind. We are also doing PSP pump storage projects. We are implementing battery energy storage systems, and there's always going to be a base load that has to come from thermal power,” Lokesh told us in an interview. While several countries have curbed new data centre developments due to resource constraints, Lokesh argued that India should take a different approach. Lokesh's pitch? Split the grid. Create dedicated renewable-powered infrastructure for data centres Green money enters the chat: Investors, too, are starting to pay attention to sustainable AI infrastructure. Vasudha Madhavan, founder of Ostara Advisors, says green capital is beginning to flow into AI infrastructure “Investors increasingly recognise that AI’s significant energy and water demands create both a challenge and an opportunity for sustainable innovation," Madhavan said. The bottom line: India's sovereign AI ambitions aren't slowing down. They can't. As Gupta of Yotta put it: “India’s sovereign AI ambition is non-negotiable. But sovereignty must go hand-in-hand with sustainability.” Check out the full story Was this newsletter forwarded to you? You can sign up for the AI Edge here
October 18, 2025
Your AI Edge Starts Here
Subscribe to read the complete newsletter and get weekly AI insights delivered to your inbox.

Subscribe Now
Free weekly newsletter
The trillion-dollar merry-go-round: Inside AI's circular economy
This week in AI, the story isn’t about a breakthrough model or some transformational use case. It’s about money...lots of money. From chipmakers to cloud giants, the AI economy is locked in a trillion-dollar loop of investments, GPU orders, and mega data-centre builds. At the centre of it all? OpenAI. Some call it visionary deal-making. Others see a bubble forming. OpenAI's web of mega-deals: Two weeks ago, Nvidia announced it would invest up to $100 billion in OpenAI. Sounds impressive, right? Plot twist: most of that money is coming straight back to Nvidia because OpenAI has committed to buying millions of their GPUs for data centres. Days later, OpenAI announced a similar tie-up with Nvidia's rival AMD, involving tens of billions in chip purchases. That deal made OpenAI one of AMD’s largest shareholders. And in September, Oracle unveiled a $300 billion partnership with OpenAI to build AI data centres, stuffed (of course) with Nvidia chips. Even CoreWeave, the Wall Street darling cloud startup, is in deep: its disclosed OpenAI compute deals top $22 billion. The numbers are absolutely wild: OpenAI has signed close to $1 trillion in compute commitments this year alone, according to The Financial Times. That's more than 20 gigawatts of capacity, equivalent to the energy output of around 20 nuclear reactors. To put this into perspective… OpenAI is valued at $500 billion but generated only $4.3 billion in revenue in the first half of the year while burning through $2.5 billion. “OpenAI is in no position to make any of these commitments,” Gil Luria, analyst at DA Davidson, told the Financial Times, who added it could lose about $10 billion this year. Even tech giants like Amazon and Alphabet have never invested capital at this scale. And those companies, you know, actually make money. How the circle works Nvidia loop Nvidia invests up to $100 billion in OpenAI OpenAI uses this cash to purchase Nvidia's AI chips OpenAI builds data centres filled with those chips Money flows back to Nvidia, now one of OpenAI's largest backers AMD connection OpenAI commits tens of billions towards buying AMD chips In exchange, AMD grants OpenAI warrants to acquire up to 10% of the company for just one cent per share If AMD's stock rises (it jumped 24% on the deal announcement), OpenAI can sell those shares to fund more chip purchases Oracle triangle OpenAI signed a $300-billion deal with Oracle to build data centres Oracle spends billions buying Nvidia chips for those facilities The money circles back to Nvidia Oracle's market value surged $244 billion when the deal was announced CoreWeave connection Nvidia holds a 7% stake in cloud provider CoreWeave Nvidia has also committed to buying $6.3 billion worth of cloud services from CoreWeave, which rents out access to Nvidia chips CoreWeave invested $350 million in OpenAI before its IPO OpenAI expanded its CoreWeave cloud contracts to as much as $22.4 billion Even Elon Musk's xAI has joined the bandwagon. Nvidia is planning to invest up to $2 billion in equity tied to a financing round where xAI will use the money to buy Nvidia processors through a special purpose vehicle. Optimists' take: Inside Big Tech, executives insist this is just what innovation looks like at scale. AMD CEO Lisa Su calls it a "virtuous, positive cycle." OpenAI president Greg Brockman argues it takes an "industry-wide effort" to meet unprecedented computing demands. CoreWeave CEO Michael Intrator told Bloomberg he's not worried about circular financing concerns. "When Microsoft comes to us to buy infrastructure to deliver to its clients who are consuming 365 or Copilot, I don't care what the narrative is. They have end users that are consuming it." Independent tech expert Tanuj Bhojwani sees particular significance in the AMD arrangement. "Specifically, the AMD deal is one where you can see OpenAI extracting a lot of value from AMD. But AMD is betting the farm on OpenAI being an anchor customer, making them relevant for everyone else," Bhojwani said. Or is this a bubble?: Sceptics see echoes of the dot-com bust. In the late 1990s, startups bought each other’s services to inflate growth. That ended... poorly. The numbers are alarming, too. Internal Oracle documents showed their cloud business generated about $900 million in sales renting Nvidia-powered servers, but earned just 14 cents gross profit per dollar, the Information reported. Moody's has flagged how much of Oracle's future depends on OpenAI and its unproven path to profitability. If OpenAI stumbles, the ripples could crash across the entire ecosystem. This all rests on one assumption: AI usage will keep growing exponentially. If that holds, today’s merry-go-round will be remembered as visionary dealmaking. But if demand slows, the whole trillion-dollar structure could wobble. For now, the merry-go-round spins faster and more expensive. Whether it delivers the future or just another bubble remains the trillion-dollar question of the decade. Was this newsletter forwarded to you? You can sign up for the AI Edge here
October 11, 2025
Bengaluru's getting an AI twin for its traffic chaos
We usually talk about artificial intelligence (AI) in terms of chatbots, copilots, or maybe even stock-trading algorithms. But here’s a real-world use case that’s about to touch millions of lives daily. Bengaluru is building an AI-powered mobility digital twin (MDT) Basically, a virtual model of the city’s traffic that learns, predicts and adapts in real time. What's a mobility digital twin, anyway?: The Bengaluru Traffic Police just floated a Rs 1 crore tender for something called a Mobility Digital Twin (MDT). Think of it as SimCity, but powered by real-world data streams, predictive AI and behavioural modelling of how we actually drive This isn't just another traffic app. The MDT will mix behavioural modelling of how people actually drive with live vehicle tracking, weather feeds, accident reports, public events and even citizen app data. It will map everything from metro stations to potholes to that random procession blocking your route home. From guesswork to machine learning: Bengaluru's traffic police aren't starting from zero. They already have tools like ASTraM (Actionable Intelligence for Sustainable Traffic Management) and traffic simulation systems. But here's the thing: these platforms sit "underutilised," according to officials, because they're not dynamic or behaviour-responsive enough Read: AI Cops! Bengaluru police pilots AI-generated avatars for social media outreach What makes this twin actually intelligent?: This isn't just a fancy mapping software. The MDT will fuse multiple data streams in real-time: Behavioural modelling of how drivers and commuters actually move (not just how they should) Live feeds from weather services, accident reports, public events and citizen apps Dynamic infrastructure mapping covering junctions, metro stations, parking zones and construction sites Enforcement databases to track repeat violators and high-risk drivers Global playbook: Globally, cities like New York, Los Angeles, Moscow, and Barcelona are already using digital twins for urban mobility. These AI-powered systems reduce commute times, simulate accident responses, and even plot optimal routes for ambulances by crunching real-time traffic data through machine learning models Bengaluru has already simulated about 3,200 km of its 14,000 km road network, covering the city's busiest corridors. Read: AI-powered signals reduce travel time by 20-33% across three road sections: Bengaluru Traffic Police The bottom line: This is AI doing what it does best: finding patterns in chaos and turning reactive systems into predictive ones. Whether it'll actually fix Bengaluru's legendary traffic remains to be seen, but at least the city's fighting back with algorithms instead of just adding more traffic cops. As one senior officer put it: “This is about moving from firefighting on the roads to foresight-driven traffic management.” Read our deep dive Was this newsletter forwarded to you? You can sign up for the AI Edge here
October 4, 2025
The great compute lockdown: How Big Tech is fortifying AI's future
Remember when the internet was an open platform where anyone with a decent computer could build the next big thing? Well, those days are officially over in AI land. We're witnessing the mother of all infrastructure grabs, with over $500 billion in deals reshaping how artificial intelligence gets built and who gets to play in the game. The result? A compute arms race that looks less like the open internet and more like a walled fortress Half a trillion and counting: In just weeks, Big Tech companies have announced over $500 billion in AI infrastructure deals. Oracle struck a record $300 billion deal with OpenAI Nvidia is preparing a $100 billion investment in the ChatGPT maker (while also backing Intel) Meta and Google tied up in a $10 billion cloud deal Tesla signed a $16.5 billion semiconductor supply agreement with Samsung This isn't just corporate dealmaking…it's infrastructure warfare. And the battlefield? Those precious GPUs, data centres, and cloud capacity that make AI magic happen. Enter Uncle Sam: Washington isn’t just cheering from the sidelines. The US government recently acquired a 9.9% stake in Intel for $8.9 billion, directly tying semiconductor self-sufficiency to AI security. Add the $52 billion CHIPS Act, and you've got Uncle Sam saying: "AI runs on American silicon, period" Why it matters: This is not just about cloud contracts; it’s a mutual dependency as a strategy. OpenAI secures guaranteed scale at predictable costs Oracle gains relevance in a market dominated by Microsoft and Amazon Nvidia locks in demand for its newest GPUs Compute power has become the rarest strategic resource in AI. Running large models isn’t possible without endless racks of GPUs, hyperscale data centres, and (let’s not forget) gigawatts of electricity. Global ripple effect: Unlike the open internet era, AI is consolidating into walled gardens. Few winners, higher costs, and centralised power While the US fortifies its AI infrastructure, the rest of the world is scrambling. China is accelerating domestic chipmaking to reduce dependence on American GPUs. The EU is floating state-backed AI funds to compete with US dominance. The Stargate project, a joint venture between OpenAI, Oracle and SoftBank planning up to $500 billion for AI data centres, perfectly captures this moment. The scale is so massive that only a handful of players can even participate. The bottom line: AI is no longer an open frontier. It’s becoming a walled garden of compute, built with half a trillion dollars, fortified by US policy and controlled by a handful of giants. The big question: does this speed us toward AGI… or lock the future of AI inside a fortress few can enter? More on this here Was this newsletter forwarded to you? You can sign up for the AI Edge here
September 26, 2025
India’s AI frenzy, public and private
Just when you thought India’s AI scene couldn’t get any louder, the government turned up the volume. While VCs can’t stop wiring money to anything with “AI” in the pitch, the government just dropped its own big bet: a fresh line-up of partners to build sovereign foundational models, including a trillion-parameter giant. Private money and public muscle are suddenly moving in sync, and the timing couldn’t be better Sovereign AI gets serious: The IndiaAI Mission has picked a new cohort to build foundational LLMs. The new players include: IIT Bombay’s BharatGen consortium, Tech Mahindra, Fractal Analytics, Avataar AI, Zeinteiq Aitech, Genloop Intelligence, NeuroDX, and Shodh AI. The headliner? IIT Bombay’s BharatGen, which will attempt a trillion-parameter model, India’s biggest yet. Backed by nearly Rs 1,000 crore in support and clusters of H100 GPUs, BharatGen will also spin out domain-specific LLMs for agriculture, governance, finance, health, law, and education Also read: IIT Bombay's BharatGen bets on trillion-parameter ‘mother model’ to power India’s AI ecosystem Fractal’s healthcare play: Fractal Analytics, meanwhile, will focus on healthcare. The company aims to build a suite of reasoning LLMs, up to 70 billion parameters in size, that can power medical diagnosis, personalised treatment, drug discovery, and optimised clinical workflows for India. It will receive 4096 H100 GPUs over nine months, with total IndiaAI support pegged at Rs 34.58 crore Also read: Fractal to release 'significant deployable reasoning AI' by year-end, once resources are in place, says CEO Srikanth Velamakanni The government’s stance: Union minister Ashwini Vaishnaw says India needs both mega-models and smaller domain-specific ones, with multiple homegrown foundational LLMs expected to be ready by the AI Impact Summit in February 2026. The government will also develop more than 500 data labs under the IndiaAI Mission to boost talent and infrastructure for artificial intelligence, Vaishnaw said Meanwhile, VCs can’t stop writing AI cheques: On the private side, things look very different, but just as frenzied. More than 12 new AI startups are prepping seed rounds. Peak XV is circling Claim Health, Cubic, Theta Software, Cua, and Coinvent AI Elevation Capital is negotiating with ScalarField, Sookti AI, and Synth Bio Kalaari, BoldCap, Zeropearl and others are chasing bets like SuperMemory, Superbryn, Sherlocks AI, and Affluense AI Most of these sub-$5 million early-stage bets are on AI agents or their development tools, with companion AI, apps designed as human-like friends or caregivers, emerging as the latest obsession. Why's everyone obsessed with AI companions? Here's where it gets interesting (and maybe a little sad?). Companion AI is absolutely exploding right now. Kavana, an AI companion platform that's building "AI humans" for conversations, has both Lightspeed and 3one4 Capital knocking on their door. The truth? People are lonely. Like, really lonely. And they're turning to AI companions for someone or something to talk to The eldercare space is particularly hot, which makes total sense when you think about it. AI companions that can provide consistent, patient interaction? That's solving a real problem. Numbers don't lie: In just the first seven months of this year, Indian AI startups raised $665 million across 109 deals. That's a 46% jump from last year's $455 million across 95 deals The average cheque size? Up 24% from $4.9 million to $6.1 million Some VCs are camping out in the US, picking startups from Y Combinator's Demo Day and signing checks on the spot (major 2021 crypto vibes). Was this newsletter forwarded to you? You can sign up for the AI Edge here
September 20, 2025

Latest AI News

Advisory Alert: It has come to our attention that certain individuals are representing themselves as affiliates of Moneycontrol and soliciting funds on the false promise of assured returns on their investments. We wish to reiterate that Moneycontrol does not solicit funds from investors and neither does it promise any assured returns. In case you are approached by anyone making such claims, please write to us at grievanceofficer@nw18.com or call on 02268882347